From: Nikodemus Siivola <tsiivola@...>
Date: 2002-12-18T11:55:12+09:00
Subject: Nature of ruby


On Wed, 18 Dec 2002, Gavin Sinclair wrote:

> Interesting.  How would the following code be compiled?
>
>   code = gets
>   eval code

Into a ruby tuby interpreter that reads it's source from the stdin. Repeat:
"Eval is just another function".

Lisp does it, so why not ruby? The problem of course is that a
minimal ruby interpreter is several orders of magnitude more complex than a
minimal ruby interpreter.

Most people on this list would be able to write something that could be
termed a Lisp interpreter in less than a week from scratch with minor help
from literature and other poeples source code -- in C, without lex and
yacc. So eval in Lisp is really "just another fucntion". You can fit "Lisp
in Lisp" on an A4.

I don't think anyone here, Matz included, would be confident in writing a
ruby interpreter from scratch in a similar timeframe.

This bring to mind another question: could you separate the "language
features" of ruby from it's "library features": what would be a minimal
subset of ruby that you would be willing to call ruby?

  -- Nikodemus